# US 11,908,124 B2 · Pavement nondestructive detection and identification method based on small samples

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 11,908,124 B2
- Document ID: US11908124B2
- Status: Granted
- Kind: B2
- Application number: 18/346,458
- Filed: 2023-07-03
- Published: 2024-01-11
- Granted: 2024-02-20
- Earliest claimed: 2022-07-05
- Assignees: Chongqing University
- Applicants: Chongqing University
- Inventors: Hui Wang; Zijie Lin; Wenruo Fan
- Web page: https://patentgrep.com/patent/US11908124B2

## CPC

- G06T 7/0004: {Industrial image inspection}
- G06T 3/4092: Image resolution transcoding, e.g. by using client-server architectures
- G06T 7/11: Region-based segmentation
- G06V 10/40: Extraction of image or video features
- G06V 10/82: using neural networks
- G06V 20/56: exterior to a vehicle by using sensors mounted on the vehicle
- G06T 2207/20021: Dividing image into blocks, subimages or windows
- G06T 2207/20081: Training; Learning
- G06T 2207/20084: Artificial neural networks [ANN]
- G06T 2207/30132: Masonry; Concrete
- Y02T 10/40: Engine management systems
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06V: IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- Y02T: CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION

## Abstract

Provided is a pavement nondestructive detection and identification method based on small samples, including: constructing an original dataset, dividing the original dataset into several patch blocks, sampling the patch blocks, and obtaining samples of the patch blocks; inputting the samples of the patch blocks into a Transformer model for feature extraction and target reconstruction, and obtaining a trained Transformer model; and based on the trained Transformer model, detecting input pavement sample images.

## Claims

### Claim 1

A pavement nondestructive and identification method based on small samples, comprising: constructing an original dataset, dividing the original dataset into several patch blocks, sampling the patch blocks, and obtaining samples of the patch blocks; inputting the samples of the patch blocks into a transformer model for feature extraction and target reconstruction, and obtaining a trained transformer model; detecting input pavement sample images based on the trained transformer model; inputting the patch blocks as an input sequence into the transformer model for training to lower resolution and reduce pixels of an image background; and performing feature extraction based on an encoder in the transformer model, comprising, firstly, obtaining the patch blocks by dividing the input equally, and then obtaining image tokens based on a method of linear projection, and adding a position embedding after the tokens are generated to solve position information lost; then, inputting labelled images into the encoder in the transformer for classification; only using class tokens in the classification; outputting different weight combinations in a process of the linear projection, wherein information obtained by the different weight combinations is multi-head information; and performing the feature extraction on the multi-head information based on a multi-head attention mechanism; the target reconstruction comprises: performing pixel-level reconstruction on masked images in the patch blocks based on an a masked autoencoder, dividing the images into patch blocks, randomly masking parts in the patch blocks, and then arranging unmasked patch blocks in sequence, and sending to a transformer encoder to obtain feature vectors; then, inserting masked patch blocks into the feature vectors according to positions in raw images, and then putting into decoder, wherein the decoder reconstructs pixel information to generate original pictures; wherein the masked patch blocks only comprise position information.

### Claim 2 (depends on 1)

The pavement nondestructive detection and identification method based on the small samples according to claim 1, wherein the original dataset is divided into the several patch blocks based on a grid splitting method to lower the resolution and reduce pixels of a picture background.

### Claim 3 (depends on 1)

The pavement nondestructive detection and identification method based on the small samples according to claim 1, wherein the encoder is a Vit encoder, and the image tokens are the class tokens.

### Claim 4 (depends on 1)

The pavement nondestructive detection and identification method based on the small samples according to claim 1, wherein an expression of multi-head attention features is:
{
Q
h
=
W
Q
h
⁢
E
T
K
h
=
W
Q
h
⁢
E
T
V
h
=
W
V
h
⁢
E
T
Z
i
h
=
∑
j
=
1
n
⁢
soft
⁢
max
⁡
(
Q
i
⁢
K
j
T
d
k
)
⁢
V
j
𝒵
h
=
[
Z
0
h
;
Z
1
h
;
…
;
Z
n
h
]
MHA
⁡
(
𝒵
)
=
W
O
⁢
𝒵
T
,
wherein n represents a number of embeddings, h is an output of an h-th head, dk represents a dimension of K, and a query Q, a key K and a value V are obtained by an embedding E and a learnable weight W.

### Claim 5 (depends on 1)

The pavement nondestructive detection and identification method based on the small samples according to claim 1, wherein verifying reliability of the transformer model based on a mean square loss (MSE) comprises: adjusting model parameters through testing TRIPLET and BCE loss, wherein a method for testing the TRIPLET and the BCE loss is:
BCE=y log x+(1−y)log(1−x)
MSE=(y−x)2
TRIPLETa,p,n=max(0,S[d(a,p)]d(a,p)−S[d(a,n)]d(a,n)+margin)),
wherein y and x are a target and prediction correspondingly, d(-) is a distance function, a, n and p are an anchor sample, a positive sample and a negative sample respectively, and S is a softmax function.